Cisco has introduced a new solution built with NVIDIA and VAST Data that aims to make enterprise data more accessible for agentic AI. At the center of this announcement is the
Cisco Secure AI Factory with NVIDIA, which now integrates
VAST InsightEngine and the NVIDIA AI Data Platform reference design. The result is an architecture designed to accelerate retrieval-augmented generation (RAG) pipelines and support enterprise-scale AI agents.
Agentic AI is about more than conversational interfaces. It’s about creating AI agents that can reason across multiple steps, collaborate, and tap into enterprise knowledge in real time. That requires fast, secure access to data - something traditional infrastructures often struggle to deliver. Cisco’s approach combines compute, networking, and data platforms to ensure AI workloads can move from proof of concept to production at scale.
Cisco AI PODs and VAST Integration
Cisco AI PODs work as the building blocks of the Secure AI Factory. With the addition of VAST InsightEngine, these PODs transform raw enterprise data into AI-ready datasets that can be accessed instantly by AI models. The architecture runs on Cisco UCS servers powered by NVIDIA RTX PRO 6000 Blackwell GPUs, providing the compute backbone for demanding AI workloads.
John Mao, VP Global Business Development at VAST Data, explained to ChannelE2E how this integration reduces friction in RAG pipelines:
“VAST’s InsightEngine is a fully integrated software and hardware solution via Cisco UCS. VAST provides purpose-built RAG pipelines leveraging NVIDIA NIMs that can analyze and understand a variety of unstructured data such as documents and videos. Our unified approach means that as data lands on the system, real-time events trigger AI pipelines automatically, invoking a chain of AI inference functions - all without human intervention. Internally, latency is further reduced because we eliminate data copying and movement through those AI pipeline stages that run within VAST’s InsightEngine.”
This integration makes Cisco one of the first vendors to deliver the NVIDIA AI Data Platform reference design in a fully validated solution. The payoff is reduced latency in RAG pipelines, giving AI agents near real-time access to business data.
Unified Architecture for AI at Scale
The Secure AI Factory with NVIDIA is built to handle the demands of agentic AI. Low-latency model interaction comes from NVIDIA accelerated computing and AI software. Cisco’s high-performance Ethernet networking ensures data and compute remain tightly connected. Security and governance are handled through Cisco AI Defense, which applies controls to every token generated.
Mao emphasized that the Secure AI Factory was designed with governance and compliance in mind:
“The first and most obvious point is that this offering can run in a private data center or co-lo of the enterprise’s choosing. None of the customer’s data leaves that premises, including all of the AI functions needed for the solution. Rather than shipping your data off to a third-party AI SaaS service, enterprises can now securely process their data directly on this platform within their complete control. Secondly, within the InsightEngine, VAST also enforces security within the enterprise’s context. For example, most enterprise data is stored with strict access controls associated. VAST leverages role-based access and user permissions by default and threads it through the AI pipelines within the system. Depending on the user querying the system, VAST’s built-in AI retrieval services only respond with answers the user is permitted to see, even though the system may know of information that spans the entire organization.”
Through this architecture, enterprises can cut RAG pipeline latency from minutes to seconds, reducing the time from query to insight; run multiple agents and workloads simultaneously, supporting continuous learning and complex reasoning; and protect sensitive data with built-in role-based access control and compliance readiness.
Enabling Partners
The Secure AI Factory is also designed with partners in mind. Cisco has made VAST’s InsightEngine software available on the Cisco GPL, giving MSPs, MSSPs, and channel partners a straightforward path to delivering this solution.
Mao described how this changes the way partners can engage. "All of the VAST software needed for InsightEngine is now on the Cisco GPL. This means any Cisco partner can easily quote and sell this VAST+Cisco solution. Because InsightEngine aims to be a true RAG ‘product,’ partners can more quickly and repeatably deploy a validated solution that works out of the box. Instead of spending time designing and configuring infrastructure and software, partners can focus on integrating the solution with custom applications or other agentic systems. It’s all about delivering quicker time to value for the customer.”
The first AI POD with VAST InsightEngine, designed for RAG acceleration, is available and Cisco plans to release a series of additional AI service PODs, each addressing new enterprise use cases. The broader implication is clear - enterprise AI adoption will depend on infrastructures that don’t just scale compute, but also manage data securely and efficiently. By working with NVIDIA and VAST, Cisco is positioning its Secure AI Factory as the foundation for that shift.
